We describe a novel process of laser-assisted fabrication of surface structures on doped oxide glasses with heights reaching 10 – 13% of the glass thickness. This effect manifests itself as a swelling of the irradiated portion of the glass, which occurs in a wide range of glass compositions. The extent of such swelling depends on the glass base composition. Doping with Fe, Ti, Co, Ce, and other transition metals allows for adjusting the absorption of the glass and maximizing the feature size. In the case of bumps grown on borosilicate glasses, we observe reversible glass swelling and the bump height can increase or decrease depending on whether the consecutive laser pulse has higher or lower energy compared with the previous one. The hypothetical mechanism includes laser heating of glass, glass melting, and directional flow. We review several potential applications of such glass swelling.
© 2009 Optical Society of America
Among the approaches for surface texturing of glass, like molding, embossing, etc., laser treatment has several advantages. Laser texturing enables easy change in pattern and the fabricating features on glasses which are difficult to mold and/or emboss. In addition, the process may be comparably fast.
Laser assisted modification of glass surfaces has been reported in many papers [e.g., 1–3] mainly using a CO2 laser. The idea of localized heating followed by fast cooling was utilized with different glasses. Since the CO2 laser wavelength (10.6 um) is strongly absorbed by most glasses, this type of laser was considered the main tool for this kind of glass processing. In addition, researchers have reported using other lasers such as the Nd:YAG or Ar-ion lasers [4–6] and the femtosecond laser . Typically these laser sources are used for producing bumps on glass surfaces with heights not exceeding several micrometers on glasses with substantial absorption levels. The concept of surface texturing using the CO2 laser is based on the local change in fictive temperature by locally heating an area above the glass transition temperature and fast cooling, leading to freezing of the non-equilibrium state of the glass.
Depending on the exposure conditions, one may control the shape of the laser-induced features. At lower pulse energies, the bump is dome-shaped, while at high pulse energies a dimple is formed .
The theory of such surface texturing has been developed for CO2 laser exposure when the laser radiation directly couples into phonon oscillations and heats the glass in the same way as a furnace would. Such laser-matter interaction allows for applying the same thermodynamic approach as when glasses go through thermal treatment in furnaces. The absorption depth in most glasses is quite small, at 10.6 μm, compared with the glass thickness (e.g., in fused silica it is ~ 10 μm ). This factor, together with glass expansion based on fictive temperature change, limits feature heights to not exceed several micrometers.
The goal of the present study was to grow bumps on glass that are 100-μm or taller. We accomplished this by choosing the appropriate base glass composition and the level of doping for optimizing laser absorption by the glass. In our experiments, we used an 810-nm pigtailed diode laser and a 1550-nm fiber laser. Unlike the CO2 laser, these lasers heat the glass through electronic absorption and, depending upon the absorption coefficient, the penetration depth can vary from several tens of micrometers to several millimeters. By using different dopants and varying their concentration in the glass, together with adjusting the laser pulse energy, we were able to maximize the extent of laser-induced glass swelling. The resulting bump heights reached 8 – 10 % of the sample thickness.
All the experiments were run at room temperature without heating the samples. The bumps were free from bubbles or cracks and they were difficult to remove from the glass without breaking the latter.
In the experiments described here, we used Pyrex-type borosilicate glasses doped with different transition metals: Fe, Ti, Co, and Ce. These dopants provide increased absorption at near-infrared (NIR) wavelengths. The doping levels were varied in order to achieve a range of absorption values and correlate them with the maximum bump heights achievable.
We used two lasers in our experiments. The first was the Spectra Physics Integra diode laser (810 nm wavelength, 60 W mean output power) with a fiber-bundle pigtail. Its emission was focused with 3-cm lens on the glass surface into a 0.6-mm spot. The intensity distribution within the spot was non-Gaussian. The second laser was a 40-W IPG Photonics 1550-nm single-mode fiber laser. For focusing, we used a 20-cm lens producing a 0.2-mm spot on the glass surface. Both lasers allowed for external modulation. Glass samples were exposed to 1-2-s pulses. At such exposures, the effect of beam shape (non-Gaussian in case of the 810-nm laser and Gaussian in case of the 1550-nm laser) does not play a significant role due to the relatively fast heat diffusion process in glass. We did not notice any difference in forming bumps as a result of change form the Gaussian beam to the multimode one. The bump height measurements were performed with a Keyence LT-9010 confocal profilometer with a resolution of better than 0.1 μm.
The dynamics of the glass exposure were analyzed using the Tardy birefringence  scheme with video camera registration. This allowed us to monitor both the dynamics of the bump formation as well as the stress development in the bulk glass. A broadband LED source was used for illumination with color filters selecting the center wavelength around 580 nm with a bandwidth of about 30 nm. The CCD camera was monochrome. Color blocking filters were placed in front of the camera to block laser radiation and heat emission from the sample. In addition, we monitored surface temperature of the glass using FLIR (forward looking infrared camera).
The evaluated glass compositions are listed in Table 1 below. Borosilicate glasses #1 – #6 are experimental melts differing mainly by Fe, Ti, and Ce content. 3966 is a Corning optical-filter glass. All listed glasses have significant boron and silica content in their compositions while the doping materials and their concentrations vary. We experimented with several ionic dopants that affect the absorption at the two laser wavelengths of 810 nm and 1550 nm and the ratio between absorptions at these wavelengths. Cerium introduces absorption mainly in the ultra-violet (UV) band. This is why optical density of glass #1 at 810 nm was quite small (0.07), with this glass being transparent at 1550 nm.
Although absorption in glasses #2 – #6 is dominated by TiO2-Fe3+ absorption, there is also a contribution from Fe2+. Glass composition, melting recipe, and heat treatment may change the absorption properties by controlling the ratio of Fe2+ and Fe3+ species. The broad absorption spectrum in these oxide glasses covers the visible and near-infrared (NIR) bands. Fe3+ leads to higher absorption in the visible and glasses look brown (glasses #2 – #6). In glasses doped with Fe2+, absorption is higher at 1550 nm than at 810 nm (3966 glass).
The thicknesses of the glass samples varied between 1–2 mm for glasses #1 – #6, to 3–4 mm in case of 3966.
Maximum bump heights and the corresponding laser pulse energies listed in Table 1 were achieved with the 810-nm laser (glasses ##1–6) and with the 1550-nm laser (3966 glass).
Such laser-induced glass swelling occurs in other glass compositions as well, which will be described elsewhere.
3. Results and discussion
3.1 Bump fabrication
Bumps were grown by focusing a laser beam on a sample surface. Laser pulse duration was usually 1–2 s and the laser-power settings were adjusted to grow bumps of different sizes.
The process of laser irradiation and of glass swelling is shown in Fig. 1. We empirically discovered that a 0.2–0.6 mm spot diameter is optimal for a 1–2 mm glass thickness. The bump forms soon after the start of laser exposure and its height somewhat decreases once the laser is switched off. The beam spot is always larger than the swelling diameter and the ratio of bump diameter to its height is around 3:1. With tighter focusing, the beam’s higher numerical aperture causes non-uniform intensity distribution in glass, which leads to a reduced glass-swelling effect. Larger beam spots lead to lower bump heights as well, probably because of the glass remelting on the surface and flowing sideways.
The propensity of each glass to form bumps was studied by measuring the bump height using a range of energy exposures from a pulsed laser. An example of the type of data obtained from these experiments is provided in Fig. 2, wherein the dependence of the bump height on the laser pulse energy with the fixed 1-s pulse duration is shown for sample #4.
We observe that for the listed glasses, bump height initially increases with increasing laser pulse energy up to a certain maximum whereat a saturation effect is observed. As the laser pulse energy continues to increase, the glass bump begins to “remelt”, and flatten out and its diameter increases. This behavior was typical for all the compositions identified in Table 1.
The maximum bump height achievable for a given laser energy pulse depends on the optical density and on glass composition. The role of glass composition was not included in this study and we focused on comparing the effect of absorption on glass swelling. From Table 1, one can see a tendency of maximum bump height and corresponding laser pulse energy to decrease with optical density (Fig. 3).
Lower glass absorption (or optical density) leads to larger absorption depth and a more uniform beam intensity distribution throughout the sample thickness. If assumed that the glass swelling effect is thermal in nature, then heating the glass more uniformly throughout its thickness is advantageous because a larger volume of glass melts and larger bumps are formed. On the other hand, heating the glass to softening/working points where its viscosity is reduced requires significantly more laser power.
In order to explore the glass response to multi-shot irradiation, we experimented with glass #4 using the 810-nm laser. First, we used 4.5-J pulse energies, targeting the middle of the acclivity observed in the representative curve shown in Fig. 2. A plot of bump height vs. the number of 4.5-J shots applied to the same spot on glass #4 is shown in Fig. 4a.
The increment in bump height is approximately 1 μm between the laser shots and the dependence on the number of shots is almost linear. Using pulse energy of 9-9.5 J, corresponding to the maximum in the representative curve in Fig. 2, consecutive irradiation still increases the bump height but the trend becomes nonlinear and saturating.
While these examples show how to increase the height of the bump the ability to lower (trim) the bumps may be equally important where a precise bump height is required. One way to accomplish this is to expose the bump to the successive laser pulses with lower energy and the results are plotted in Fig. 4b.
The first shot created a bump about 112 μm tall. The next two shots at 4.5-J pulse energy brought this height down to 110.7 μm. These latter shots reduced the height by approximately 0.5 μm each.
Such reversible glass swelling is observed on the linear ascending slope of the curve in Fig. 2. If the starting bump height is more than, say, 114 μm, then the bump height cannot be reduced.
Based on the experiments with increasing and decreasing the bump height with consecutive exposures, we studied the feasibility of precise bump-height adjustment. One way of doing it is to expose the glass with laser pulse energy taken from Fig. 2. However, due to variation of glass properties within the sample and laser-pulse instability, we were unable to achieve the target height with better than several micrometer accuracy. The adaptive process illustrated by Fig. 5 helped to significantly improve the accuracy of bump adjustment.
Glass #5, when irradiated with a 1550-nm laser, had a response curve similar to that shown in Fig. 2. We selected an arbitrary target bump height of 70 μm with a required precision of 0.1-μm. Such precision was in line with the accuracy of the Keyence LT-9010 confocal profilometer. The predicted laser pulse energy was 4.25 J but the bump height achieved was actually 1.6 μm short of the target. Shooting with a pulse of higher or lower energy, depending on the deviation of the actual bump height from the target, enabled us to adjust the height to 70.1 μm, which was in the required range.
Similar iterations were carried out when the initial bump height was several micrometers above the target. In this case, lower pulse energies were used to reduce the height. The range of adjustment depends on where the target height is on the slope of the characteristic curve. In the upward direction the height is limited by the described saturation effect. To decrease the bump height, one must be within the trim range of 5–10 μm.
This remarkable characteristic of reversible swelling occurs in other boron-containing oxide glasses like 3966 commercial glass.
In addition to bumps, raised ridge structures may be formed on IR-absorbing glass substrates by sweeping a focused laser beam along the substrate surface. For example, ridges up to 47 μm high have been created using a ~200 μm diameter 1550-nm laser beam (6 W) sweeping at 1.0 mm/s. These ridges have been formed along straight paths, curved paths and through right angle bends. Raised features can be fabricated in arbitrary locations for precision alignment of components, or alternatively closed ridge features can form wells or microchannels to constrain working fluids or adhesives.
Ridge height can be tuned by adjusting laser power and sweep velocity. Fig. 6 shows measurements of glass laser ridges as a function of laser power (2.5 – 6.5 W) for two different sweep velocities (0.25 mm/s and 1.0 mm/s). Ridge height increases with higher laser power up to 6 W and then decreases. Ridge height increases as sweep velocity decreases.
Ridge height varies slightly along the length of the ridge due to non-uniform experimentally melted glass composition. Measurements of ridge heights at several positions along a ridge show that typical ridge height variation is below ±2.0 μm. The width of the ridge also increases with increased laser power and reduced sweep velocity, varying from ~100 to ~400 μm for the cases covered in Fig. 6.
The bump growth process on 3966 glass was studied using an on-line birefringence setup. We used the Tardy scheme  with two quarter-wave plates and two crossed polarizers. The dynamics of a 3-mm thick glass exposure process with a 2-s, 810-nm laser dose is shown in Fig. 7. The peak stress in the sample after exposure is close to 70 MPa. This dynamic shows that significant volume inside the glass is involved in the bump formation. The stressed (bright) areas in the photos are of the affected glass. The expansion of the glass due to heating increases over time and, at a certain point, the glass starts to flow. This is seen in the stress patterns at 1.6 s and 2 s after exposure start where visibly brighter division lines appear. This flowing-glass region increases over time. Actual size of the bump formed at the end of exposure is significantly smaller that heat affected area and is not visible in the photos. The bright spot on the surface of the glass seen at 2.0 s after exposure is an artifact of measurement and is attributed to the hot-glass emission at wavelengths close to the ones at which birefringence was measured.
We observed in the Tardy setup arrangement with the polarizers removed that there is a noticeable refractive index difference between the glass affected by the heating and the surrounding glass. The measurements of refractive index values showed reduction in the refractive index close to 0.2 % of the initial value indicating the decrease of glass density. This is consistent with the fictive temperature model predicting ~ 0.1–0.2 % change in glass volume.
The dynamics of the bump growth process on 3966 glass with the 810-nm laser were also studied in a separate experiment. The bump height and its corresponding peak temperature as a function of time (measured with FLIR camera) are shown in Fig. 8. The bump becomes noticeable when temperature exceeds 600–800 °C, which is above the softening point of the glass. After the glass cools to a temperature below its softening point, the bump shape freezes. Some reduction in bump height occurs during the ~0.1-s quenching time due to thermal contraction of the glass right after the laser exposure.
A simplified explanation for bump formation is proffered as follows. After the laser begins to heat the glass, the temperature around the beam axis elevates beyond the softening point and liquid glass flows and expands upward. The surrounding laser-heated volume remains solid but expands inwards because its expansion outside into the cooler bulk of the glass is restricted. This expansion “squeezes” the liquid glass up further, forming the bump (the bump shape is spherical due to surface tension). After the laser pulse ends, the surface of the bump quenches faster than the volume below it due to radiative heat loss, which effectively “freezes in” the bump shape. The glass under the bump quenches under fixed-volume restriction, which results in tensile stress.
As seen in Fig. 7, the process is linked to residual stress in the glass due to fast heating and quenching, resulting in noticeable birefringence. This stress can be partially annealed by heat treatment in the furnace with very small modification of the bump shape. Typical shrinkage is about 1–2 um in height.
We do not yet fully understand the mechanism of bump regrowth, which is probably associated with heating the glass to higher or lower temperatures depending on the laser pulse energy.
3.3 Lens fabrication
The bump growth process in glass can be used to fabricate lenses or lens arrays. We monitored the parameters of the created surface, which is spherical due to the surface tension (Fig. 9).
The curvature radius can be modified in a wide range from 80 μm to at least 600 μm, which can be achieved by varying either laser exposure parameters, or the laser spot size on the target, or both. The lens diameter ranges typically from 200 μm to almost 1 mm. As shown in Fig. 10, the radius and diameter of the lens changes somewhat in concert.
The focusing properties of these lenses are determined mostly by the bump shape. The refraction caused by the small index decrease of ~ 0.1–02 % in the glass below the bump is negligible compared to the refractive properties of the bump itself.
This wide range of the lens curvatures and of their diameters provides a significant opportunity to alter the lens parameters. The numerical aperture (NA) of lenses fabricated in this way can vary from less than 0.2 up to over 0.45. If needed, these lenses can be spaced very close together with center to center spacing less than 1 mm as presented in Fig. 11.
Despite this close proximity of these lenses, we did not observe any glass damage in the lenses or in the glass beneath them. The NAs of the individual lenses in the array differ by several percent and this variance can be further reduced by bump regrowth in the manner described above.
This lens and lens array fabrication approach is an advantageous alternative to molding allowing for flexibility in changing the lens parameters and for high placement precision and low NA variance.
We demonstrated significant (extraordinary) glass swelling (in excess of 10% of the glass thickness) in doped glasses with near-IR lasers. In the borosilicate glasses that we experimented with, we were able to adjust the bump height with a sub-micron precision by iteratively applying higher or lower laser exposure.
As an example of a practical application, this laser-induced glass swelling effect allows for fabricating microlenses and their tightly packed arrays with NAs ranging from 0.2 to about 0.5. We believe that the described phenomenon can find its way into other novel applications. Furthermore, these structures can be used to perform an extremely precise out-of-plane alignment of micro-components.
The authors are grateful to T. J. Kiczenski for creating glass compositions #1 – #6 and to J. Dickinson and D. Allan for fruitful discussions.
References and Links
1. T.D. Bennett and L. Li, “Modeling laser texturing of silicate glass,” J. Appl. Phys. 89, 942–950 (2001). [CrossRef]
2. T.-R. Shiu, C. P. Grigoropoulos, D. G. Cahill, and R. Greif, “Mechanism of bump formation on glass substrates during laser texturing,” J. Appl. Phys. 86, 1311–1316 (1999). [CrossRef]
3. V.P. Veiko and E.Y. Yakovlev, ‘Physical fundamentals of laser forming of micro-optical components,” Opt. Eng. 33, 3567–3571 (1994). [CrossRef]
4. T.H. Kim, K.H. Lee, Y.J. Jung, Y.S. Kim, H.J. Chin, B.K. Ryu, H.O. Kang, and S.H. Lee, “Micromachining of glass by using a Nd:YAG (532 nm) laser for the optical device.” J. Ceram. Soc. Jap. 116, 309–312 (2008). [CrossRef]
6. M. Fritze, M.B. Stern, and P.W. Wyatt “Laser-fabricated glass microlens arrays,” Opt. Lett. , 23, 141–143 (1998). [CrossRef]
7. Y. Hayasaki and D. Kawamura, “High-density bump formation on a glass surface using femtosecond laser processing in water,” Appl. Phys. A 87, 691–695 (2007). [CrossRef]
9. MLH. Tardy, “An Experimental Method for Measuring the Birefringence in Optical Materials,” Opt. Rev. 8, 59–69 (1929).